data logo

Known bugs in current and previous versions of the Gemini IRAF Package

Known bugs (and workarounds, where possible) in the Gemini IRAF package are described here. If you find a new problem, please submit the information via the Gemini HelpDesk.

6 November 2006
FITCOORDS bug in IRAF 2.13beta2

There is a known bug in IRAF 2.13beta2 FITCOORDS. This will affect users of the Gemini package who are reducing spectroscopy data. At this time we know that the problem shows up during the reduction of NIFS data. It is likely to affect the reduction of the other spectroscopy data at the fitcoords/transform step.

We recommend that you continue running the Gemini package with IRAF 2.12.2a.

For the Mac Intel users who have to use IRAF 2.13beta2, we suggest you download and install the unofficial patch below. The fixed code was graciously provided by the IRAF group. The binaries were compiled at Gemini. If you encounter problems with it, contact Gemini.

fitcoords213b2-macintel.tar.gz
fitcoords213b2.README

20 February 2006
GNIRS - 2005B and on : two new database files are needed (Files included in version 1.9 and up)

Due to a change in the GNIRS cameras, two new database files are needed for reducing 2005B and on GNIRS short camera data with the Gemini/GNIRS IRAF package (v1.8.1). More information is available in an email sent to the GNIRS PIs, and the two files are available below. The updated files will be included in the next IRAF release.

21 September 2005
Problem with GSFLAT when fl_detec=no and more than one input flats (version 1.7 and 1.8; fixed in subsequent versions)

When several flats are used as inputs, GSFLAT uses only the first input flat in the list rather than a combined frame to derive the output flat. This only happens when fl_detec=no; the fl_detec=yes is not affected by this bug. The bug affects GSFLAT in Gemini v1.7 and v1.8. Patches have been made available for both 1.7 and 1.8 versions of the Gemini package:

VersionPatch DownloadInstructions
1.7 gemini_v17_patch2.tar.gz gemini_v17_patch2.txt
1.8 gemini_v18_patch1.tar.gz gemini_v18_patch1.txt

10 November 2004
nsprepare crashes with bad WCS or fl_checkwcs- fl_forcewcs+ (version 1.7, fixed in subsequent versions):

The task nsprepare will crash if fl_checkwcs=yes (the default) and the WCS matrix has a negative determinant. This can be avoided by using fl_checkwcs=no. However, this then means that fl_forcewcs=yes cannot be used to correct the WCS. To use fl_forcewcs=yes and fl_checkwcs=yes together you must zero the values of the CD_ij elements in the PHU (using hedit) which will allow fl_checkwcs+ to be used without crashing (and so allow fl_forcewcs to be used too). Alternatively, you can download this nsprepare.cl file and replace the version of it that came with v1.7.

4 November 2004
nireduce crashing when fl_autosky=no (version 1.7, prior to November 8):

The task nireduce will invariably crash if fl_autosky is set to 'no'. This problem is caused by a typo that has been introduced in v1.7. Previous versions are not affected. You can download and install this patch to remedy the problem (instructions). If you have downloaded the v1.7 software from the Gemini web site after November 8, the new version of nireduce.cl is already included and you do not need to apply the patch. To verify the version that you are currently running start the CL, load the gemini package, and type print(gemini.version); if the version date is '25Oct2004', it is recommended that you install the patch.

4 November 2004
giflat and gifringe having problems with data binned 4 by 4 (version 1.7, prior to November 8):

The tasks giflat and gifringe do not set the default 'statsec' and 'normsec' correctly for GMOS data binned 4 by 4. You can download and install this patch to remedy the problem (instructions). If you have downloaded the v1.7 software from the Gemini web site after November 8, the new giflat.cl and gifringe.cl are already included and you do not need to apply the patch. To verify the version that you are currently running start the CL, load the gemini package, and type print(gemini.version); if the version date is '25Oct2004', it is recommended that you install the patch.

5 June 2003
gmosaic and binned images (version 1.4, prior to June 5, 2003, and version 1.3):

The gmosaic task to mosaicing the GMOS-N detectors does not correctly handle binned images. The bug affects images that are binned in X as well as images that have different binning in X and Y. For images binned in X, chip 1 and 3 were placed too far from chip 2 in the mosaiced image. For images with different binning in X and Y, the rotations of chip 1 and 3 relative to chip 2 was incorrect. Processing of multiple images with binning in X was also implemented incorrectly. The bugs have been fixed. In addition, the mapping of chip 3 relative to the other chips has been improved. If you have installed the Gemini IRAF package v1.4 prior to June 5, 2003, you can download a new gmosaic.cl file and replace the old version. You will also need new versions of the bad pixel masks included with the software: mgmos_bpm11.pl and mgmos_bpm22.pl. These files should be copied to the directory "gmos$data/" If you have downloaded the software from the Gemini web site after June 5, 2003, the new versions of gmosaic and the bad pixel masks are included.

Should I rereduce my data? Imaging data with small dither steps show no significant effect. Unless you are doing high precision astrometry there is no reason to rereduce imaging data. For spectroscopic data the answer to the question depends on your science. The effect of the bug on data binned by two in the X-direction is a misplacement of chip 1 and 3 relative to chip 2 by about two unbinned pixels. If you are measuring high precision radial velocities or measuring precise linewidths by fitting template spectra to a large spectral range you may want to rerun part of the reductions for testing purposes. However, for measurements of redshifts the effect is too small to cause any significant effects on the results.


5 June 2003
gnsskysub task missing (version 1.4 only, prior to June 5, 2003):

The task gnsskysub for reduction of Nod-and-Shuffle data from GMOS is missing in the Gemini IRAF package v1.4. If you have installed the Gemini IRAF package v1.4 prior to June 5, 2003, you can download the gnsskysub.cl file and the help page and copy them to the directories "gmos$" and "gmos$doc/" in your installation. If you have downloaded the software from the Gemini web site after June 5, 2003, gnsskysub is included.

5 June 2003
gireduce unstable under Linux/Redhat (version 1.4 prior to June 5, 2003, and version 1.3):

It has been found that the use of the keypar task to access the image headers is not stable under Linux/Redhat. gireduce has been updated to use imgets instead. Other tasks affected by this will be updated in a future release of the Gemini IRAF package. If you have installed the Gemini IRAF package v1.4 prior to June 5, 2003, you can download a new gireduce.cl file and replace the old version. If you have downloaded the software from the Gemini web site after June 5, 2003, the new version of gireduce is included.

8 October 2002
mkpkg file missing (version v1.4 only, prior to Oct 8, 2002):

The top level mkpkg file was missing in the released version v1.4. If you have downloaded that version and you need to compile the software, you can download the mkpkg file here. If you have downloaded the software from the Gemini web site after Oct 8, 2002, the mkpkg file was included, and you do not need to download this file separately. The file is also not needed if you do not need to compile the software. Compilation is not needed for Solaris/UNIX and Linux operating systems.

25 September 2002
imcoadd (version 1.3 only) - incorrect use of limit parameter:

In the development of the Gemini IRAF package version 1.4 it was found that imcoadd released with version 1.3 has a bug in the cosmic ray cleaning. The use of the limit parameter in the case where this is given as a sigma limit above background was not implemented correctly. Only for the first image in the stack was the implementation correct. For the remaining images in a stack the limit in counts was set too high. The result is that the cosmic ray cleaning is done too close to the cores of the objects. For bright objects and images with differences in the image quality, this can result in compromised photometry. The bug is fixed in version 1.4. The bug was not present in releases earlier than v1.3.

25 September 2002
Missing header keywords (version 1.4 and earlier):

The Gemini IRAF package relies on the header keywords for many of the processing steps. Occasionally, the various tasks will fail to update the headers of intermediate processing steps correctly. As a result, later steps will fail. The behaviour is intermittent and has been reported most often under various versions of Linux. In some cases, the problem can be solved by re-reducing the images for which the reduction failed the first time around. In other cases, it has been reported that manually editing the headers of the intermediate results and then proceeding was a possible work around. It is possible that the underlying problem is in the core IRAF. When the Gemini IRAF package is ported to IRAF 2.12, we will document the problem in more detail and work with the NOAO/IRAF group to get the problem solved.

25 September 2002
MDFs not found in images (gmos package version 1.4 and earlier):

Occasionally, IRAF will fail attach an MDF to image even though the required flags are set correctly and the MDF exists. As a result, later steps will fail. The behaviour is intermittent and has been reported most often under various versions of Linux. The problem can be worked around by manually re-attaching the MDF using the task fitsutil.fxinsert. It is possible that the underlying problem is in the core IRAF. When the Gemini IRAF package is ported to IRAF 2.12, we will document the problem in more detail and work with the NOAO/IRAF group to get the problem solved.

[Science Operations home][Data home]


In original form; Inger Jørgensen
Last update November 6, 2006; Kathleen Labrie